OctoMed: Data Recipes for State-of-the-Art Multimodal Medical Reasoning
This work addresses the need for high-quality data to improve generalization in medical AI systems, though it is incremental as it builds on existing supervised fine-tuning methods.
The authors tackled the problem of training robust multimodal medical reasoning models by investigating data curation strategies, achieving state-of-the-art performance among open-source models on diverse out-of-distribution medical benchmarks with a dataset of over 8 million examples and 6.8 billion response tokens.
High-quality and carefully curated data is a cornerstone of training medical large language models, as it directly impacts both generalization and robustness to unseen clinical tasks. We investigate strategies for training and data curation to develop a robust multimodal reasoning model in the medical domain. Our work focuses on supervised fine-tuning (SFT) and explores data recipes that leverage structured reasoning traces. Using our proposed data recipe, we scale experiments to a dataset of over 8 million examples and 6.8 billion response tokens, achieving state-of-the-art performance among open-source models across diverse out-of-distribution medical benchmark tasks. Our results further indicate that curating a high-quality, diverse training dataset with varying structured reasoning trace lengths enables the fine-tuned model to self-calibrate its reasoning trajectory lengths based on the downstream task, without explicit supervision. We present key insights, describe the data curation strategy, and outline next steps toward developing robust medical vision-language reasoning system.